Unsupervised Learning:

How it works:
Unsupervised Learning is a type of Machine Learning where algorithms analyse data without any labelled outcomes or predefined categories.
Instead of being told what to look for, the system tries to discover patterns, groupings, or structures on its own.

Anomaly Detection:
Unsupervised Learning is particularly useful for Anomaly Detection.
In Finance: Fraudulent transactions, Accounting irregularities, Abnormal customer behaviour.
In Cybersecurity: Network intrusions, Unusual login behaviour, Malware or bot activity.
In Industry: Sensor malfunctions, Machine vibration anomalies, Unexpected temperature or pressure changes.

Customer Segmentation:
Profiling customers behaviour patterns.
Market basket analysis.

Identifying Representative Weather Patterns:
Discover hidden variables in climate data.
Grouping climate zones.

Unsupervised Learning can be broken down into three phases:
Data Preparation Phase: Cleaning, normalising, scaling, selecting features, removing noise and irrelevant variables.
Pattern Discovery Phase: The algorithm learns structure from the data it identifies groups, latent features, or patterns without labels
Interpretation & Deployment Phase: Interpreting clusters, embeddings, or anomalies.

Unsupervised Learning uses five Algorithms:
Clustering: Grouping data points based on similarity.
Dimensionality Reduction: Compressing data while preserving structure.
Association Rule Learning: Discovering relationships between variables in large datasets.
Density Estimation: Modelling the underlying probability distribution of data.
Anomaly Detection: Identifying anomalies built on clustering or density estimation.